The Basel Action Network (BAN), a nonprofit organization, estimates that data centers built to serve artificial intelligence could produce the equivalent of 23 million shipping containers of electronic waste by 2050. That projection is substantially higher than many previous estimates because BAN included more than just servers in its calculations.
What was counted
BAN researchers included supporting infrastructure items that often get left out of headline figures: power supplies, cooling systems, uninterruptible power and backup systems, and networking equipment. According to BAN, tallying these components provides a more complete picture of how much e‑waste the AI sector will leave behind.
Current e‑waste and future projections
Today the world generates about 68.3 million tonnes of electronic waste annually, and less than a quarter of that is recycled. BAN projects that global e‑waste could rise to 211 million tonnes per year by 2050, and that 15–20 percent of this volume could originate from AI data centers alone.
Warning and potential consequences
"AI may seem weightless, but every model runs on a lot of highly specialized, top‑tier hardware," said Jim Puckett, founder of the Basel Action Network. Puckett cautioned that if governments and companies do not prepare for this coming "waste tsunami," the build‑out of AI infrastructure could create problems more severe than the present e‑waste situation.
Local opposition and energy concerns
Resistance to AI data center projects has been growing. In some places local officials were voted out over such projects; New York imposed a moratorium on building data centers; and in one Texas city controversy arose when land originally intended for a public park was sold to a data center developer. Policymakers in Texas have also raised concerns about the sector's high energy consumption, and some local leaders considering data center projects have reportedly received death threats.
Why this matters
Expanding AI services brings technological and economic benefits, but the hardware demands, increased resource use, and resulting electronic waste could create significant environmental and social costs unless accompanied by regulation, recycling capacity, and transparent planning. BAN's report emphasizes that policymakers should assess the full supply chain — not just servers — when developing strategies to make AI infrastructure more sustainable.
